Printer Friendly

EVOLUTION OF CLUTCH SIZE IN THE EASTERN KINGBIRD: TESTS OF ALTERNATIVE HYPOTHESES.

MICHAEL T. MURPHY [1]

Abstract. David Lack proposed that parental feeding ability ultimately limited clutch size in bird species in which the young were dependent upon their parents for food. However, many species can raise broods that are larger than their normal clutch size. Based on nine years of experimental results from an individually marked population of Eastern Kingbirds (Tyrannus tyrannus) breeding in central New York (USA), I test six hypotheses that have been proposed as explanations for why birds fail to lay larger, seemingly more productive clutch sizes. I modified brood sizes by adding or removing 1-2 nestlings when broods were 1-3 d old and then documented the effects of brood size and manipulated brood size on nestling size and survivorship, offspring recruitment, adult survival, and future adult reproduction. Most first clutches of the season held three eggs (62% of 503 clutches), but the proportion of young to fledge did not vary with brood size (1-5 young), and as a result, broods of five were the most producti ve. Lack's basic food-limitation model was thus rejected. Although nestling mass and ninth-primary length at fledging declined with brood size, offspring survival during the immediate 10-12 d period after fledging was unrelated to nestling mass or lengths of the tarsus or ninth primary. The findings that the underweight young in broods of four and five did not suffer disproportionate mortality and that they were just as likely to appear as recruits in future years led to a rejection of the extended version of Lack's food-limitation model.

Comparisons of annual variation in the relationship between productivity and brood size showed that productivity increased with brood size in eight of nine years (significant in six years). Thus, high temporal stochastic variation in conditions for rearing young (the "bad-years" hypothesis) is unlikely to explain the relatively small clutch size of kingbirds. Predictions of two other hypotheses that predict asymmetrically low survivorship of young in large broods (the "cliff-edge" and "brood-parasitism" hypotheses) were also rejected. On the other hand, evidence suggested that females individualize clutch size such that each female lays a clutch that matches her individual feeding ability.

Although fledgling production was not adversely affected by experimental increases in brood size, most enlarged broods lost young during the 10-12 d immediately after fledging. Thus, enlarged broods ultimately produced no more independent young than did control broods that began with the same number of eggs. Fledgling deaths were not related to nestling mass or size, and recruitment was independent of manipulations. Survival and fecundity costs of reproduction also existed for females. Male survival (68%) was independent of the number of young that had been raised (0-5 young), and future breeding efforts were not compromised by elevated effort in the past year. However, females that raised broods of five were less likely to return to breed in the following year (42%) than were females that raised 2-4 young (62%). Among the survivors, females that raised enlarged broods in the preceding year also experienced more hatching failure and fledged fewer young than females that raised reduced broods in the preceding year. I suggest that costs of reproduction probably set the ultimate limit to clutch size in Eastern Kingbirds. I did not test the hypothesis that high rates of nest predation favor the evolution of small clutch size, but given that predators destroyed [tilde]50% of nests each year, it is also likely that nest predation has contributed to the evolution of the current clutch size of kingbirds. Whether a female produces a clutch of three or four eggs is probably determined by individual differences in parental ability, which may be related to either intrinsic properties of the female or territory quality.

Key words: brood size; clutch size, evolution of; fecundity, future; individual optimization; life-history theory; nest predation; offspring recruitment of brood size, fledging size; passerines; reproduction, cost of; survival; Tyrannus tyrannus.

INTRODUCTION

Life-history theory (Stearns 1992) and most models of the evolution of clutch size in birds (see below) and other organisms (e.g., Godfray 1987, Lloyd 1987, Morris 1992, Risch et al. 1996) have been heavily influenced by Lack's (1947) argument that altricial birds lay the number of eggs that correspond to the maximum number of young that can be fed. His prediction has been tested experimentally numerous times and largely refuted because of the demonstrated ability of parents of many species to fledge young of normal size from enlarged broods (reviewed by Dijkstra et al. [1990], Stearns [1992], and VanderWerf [1992]). The reason why many species do not conform to the "Lack clutch size" (sensu Godfray et al. 1992) is the subject of long-standing debate (Monaghan and Nager 1997).

To account for the discrepancy between theoretical predictions and empirical findings, Lack suggested that young from successful broods that were larger than the modal clutch size were underfed, small, and less likely to survive to adulthood (Lack 1966). In other words, an intergenerational cost of reproduction existed because parents passed the cost onto their offspring. A number of species do show evidence of low survivorship of young that fledge from large broods (Tinbergen and Boerlijst 1990, reviewed by Magrath [1991]), but the pattern is certainly not universal (e.g., Hochachka and Smith 1991). Alternatively, Skutch (1949) and more recent authors (Slagsvold 1982, Lima 1987, Martin 1992) argued that frequent nest predation favors broods that are smaller than the maximum number that can be fed. The rationale varies slightly among authors, but the substance of the argument is that large broods are more likely to attract predators, increase predation risks for parents or reduce the time that parents have t o guard the nest, and as a result, large broods fledge few young. Williams (1966; see also Charnov and Krebs 1974) then proposed that adults optimize current effort and future reproductive value to maximize lifetime reproductive success. The essence of Williams' widely accepted proposal is that adults do not jeopardize either their survival or future fecundity for immediate reproductive gain. The intragenerational "cost hypothesis" is a pillar of life-history theory, and abundant documentation of phenotypic costs of reproduction exist from field studies of birds (e.g., Gustafsson and Sutherland 1988, Stearns 1992, Daan et al. 1996, but see Reznick [1985, 1992] for comments on relevance of field measurements of phenotypic costs). Theory predicts that reduced breeding effort is likely to evolve when costs exist and juvenile mortality exceeds adult mortality (Law 1979, Reznick et al. 1990). Hence, selection for small clutch size may be most intense when nests are frequently lost to predators, adults have potenti ally high survival, but reproduction is costly (e.g., Julliard et al. 1997).

Three additional adaptive hypotheses that all rely upon uncertainty in the success of large broods have also been proposed to explain why females do not lay larger clutches. Mountford's (1968) "cliff-edge" hypothesis suggests that selection favors broods that are smaller than the long-term optimum if females cannot produce a clutch (or litter) of an exact size, and those females that overshoot the optimum produce very few surviving young. Mountford argued that the asymmetrically low survivorship of young from broods above the optimum favored individuals that left a margin of error by producing smaller clutches (see also Morris 1992, 1996). The "parasitism-insurance" hypothesis (Power et al. 1989) also depends on asymmetrically low survivorship of young from large broods, but in this case, the unpredictable laying habits of brood parasitic females enlarges clutches so that the pair can no longer feed the brood adequately. To safeguard against low reproductive success (and possibly total failure) of parasitize d broods, females presumably leave space in the clutch as "insurance" against the possibility of receiving eggs from other females. The "bad-years" hypothesis (Boyce and Perrins 1987) proposes that females that lay clutches that are smaller than the apparent optimum have the highest fitness when optimum brood size varies stochastically across years. The basic premise is that, although smaller broods may often produce fewer young than large broods, high variance in the success of large broods ultimately lowers their fitness below that of individuals that reproduce more conservatively (Murphy 1968, Gillespie 1977, Frank and Slatkin 1991; but see Liou et al. 1993, Cooch and Ricklefs 1994). Although modeled originally for haploid organisms with nonoverlapping generations, strategies that minimize variance in breeding success increase mean fitness (Seger and Brockman 1987) and the principle may well apply to diploid organisms with overlapping generations.

Nonadaptive explanations for the discrepancy between modal clutch size and most-productive brood size also exist. For instance, given that birds often produce clutches prior to the peak in seasonal food abundance (e.g., Murphy 1986), food availability may limit egg production. Although not unreasonable, it is doubtful that food supply directly limits clutch sizes of small, altricial birds because most eggs are produced from daily food intake (Winkler and Allen 1995, Perrins 1996) and egg-production costs are only a small fraction of existence metabolism (e.g., Ward 1996). Furthermore, food and nutrient supplementation rarely results in larger clutches (Meijer et al. 1990, Ramsay and Houston 1997; reviewed by Nager et al. [1997]). Finally, females might individualize clutch size to match either their personal condition or territory quality (Lack 1966, Perrins and Moss 1975). In essence, each female has her own "Lack clutch size." In this case, productivity is expected to increase linearly with brood size in u nmanipulated nests because each female has laid the number of eggs that matches her expected parental ability. Productivity is predicted to level off or even decline when broods are enlarged (Hogstedt 1980, Pettifor et al. 1988, Pettifor 1993a).

Despite decades of efforts, remarkably few species have been studied well enough to establish why many bird species fail to lay what would appear to be larger, more productive clutches. Part of the reason for the slow progress has been the highly selective nature of most tests. Only rarely have more than one or two of the hypotheses that I have listed been tested when a species is known to be able to raise enlarged broods. Platt (1964) reasoned that the simultaneous testing of multiple working hypotheses was the most efficient approach for resolving questions with multiple potential answers. To this end, in this report I demonstrate that Eastern Kingbirds (Tyrannus tyrannus) can raise broods that are 67% larger than the modal clutch size, and test six hypotheses (Table 1) that have been proposed as explanations for the failure of birds to lay larger, seemingly more productive clutches. My results indicate that the upper limit to clutch size in Eastern Kingbirds is probably set by intragenerational costs of r eproduction (and possibly nest predation), but that intrapopulation variation in clutch size is tied to individual or site-specific differences in parental ability.

NATURAL HISTORY

Eastern Kingbirds arrive in North America from their wintering grounds in South America by late April/early May. Almost all surviving males reuse former territories (Murphy 1996a), and pairs form socially monogamous pair bonds by mid- to late May. Most females reestablish pair bonds with former mates (Murphy 1996a). Females then build open-cup nests that are placed in trees at a height of 2-8 m (Murphy et al. 1997). Modal clutch size is three or four eggs, depending upon the populations (Davis 1955, Murphy 1983a, Blancher and Robertson 1985). Females incubate (14-16 d), and then both sexes feed the young for the 16-17 d nestling period. Females feed the young more than males, but males spend more time guarding the nest and chasing potential nest predators (Woodard and Murphy 1999). The young then remain with the adults for 3-5 wk after fledging (Morehouse and Brewer 1968). Only one brood is raised annually, but most females make at least one attempt to replace a failed nest before southward migration begins in August (Murphy 1996b).

METHODS

Detailed descriptions of field procedures (Murphy 1983a) and study site (Murphy 1996a) are given elsewhere. I have studied the Charlotte Valley population of Eastern Kingbirds from Delaware County, New York (USA; 42[degrees]27' N, 74[degrees]47'W) since 1989, and in this report I include all data collected from 1989 through 1997, plus information on recruitment and adult survivorship up through 1998. In all years suitable sites and all territories that were used previously were intensively checked for returning birds beginning in mid-May. I made repeated visits to determine whether potential breeding sites were used, and then the following breeding statistics were collected for all breeding pairs: breeding date (date of first egg), clutch size (number of eggs/nest), mean egg mass, number of eggs to hatch, number of addled eggs, brood size (number of young/nest), number of young to fledge, and nestling size at 13 d after hatching ([plus or minus]1 d). Nests were checked every 3-5 d during the nestling period to determine if the nest was active or if individual young were missing. Most nests were found prior to egg laying. Thus, to calculate nest success I divided the number of successful nests (= fledged at least one nestling) by the total number of nests. Almost all nest failures were due to nest predation or starvation, and I assumed that losses of entire broods or individual nestlings within nests in which the remaining young showed no signs of food deprivation were the result of predation. Death was ascribed to starvation if young were found dead in the nest or individual nestlings were missing from broods showing evidence of poor nutrition.

At day 13 young were measured and banded with one U.S. Fish and Wildlife Service band and three plastic colored leg bands to permit individual identification. Nestling measurements included mass (in grams), wing chord (in millimeters), and lengths (in millimeters) of the tarsometatarsus (= tarsus) and bill (1991-1997 only). I also attempted to capture the parents using mist nets just after day 13. Adults were individually color banded and standard measurements were taken (mass, wing chord, and tarsus and bill length). Blood samples were collected from nestlings and adults at some nests to test for extra-pair parentage. If nests failed I determined whether or not they were replaced, and then collected the same data for all renests, but in this report I focus solely on initial nests of the season. From 1991 through 1994 I also documented fledgling survival for a subsample of nests during the 2-3 wk period after fledging. I returned to territories when the young were between 28 and 35 days of age to count the n umber of adults and fledglings. If no birds were found I assumed that the family unit had moved off the territory. Finally, I measured adult and nestling return rate to the population by observing or capturing previously banded birds (Murphy 1996a).

In eight of nine years I manipulated brood sizes within 1-3 d of hatching by adding or removing 1-2 young of the same age ([plus or minus] 1 day) as the young in the nest. Age differences of 1-2 d are normal for kingbird nestlings since most clutches hatch over a 24-36 h period (Murphy 1983a). I selected nests for manipulation on the basis of proximity, overlap in hatching date, and same clutch size. I attempted to designate three broods as control (brood size = clutch size), enlarged and reduced, but it was often impossible to find three broods to fit these criteria because kingbird nest density is low (Murphy 1996b) and nest failures desynchronize nesting.

Statistical analyses

I used analysis of variance (ANOVA) to determine if brood size negatively affected productivity (i.e., number of fledged young), nest success (proportion of nests to fledge young), fledging success (proportion of young to fledge), and nestling quality (i.e., mass and size). Sample sizes vary among analyses because not all measurements were taken in all years (e.g., nestling bill length), and samples of nestling size were smaller than for other variables because I could not measure young in some nests that were placed out of my reach. Proportions were arcsine transformed, and all variables were tested for equality of variances. If group variances were unequal I used the Kruskal-Wallis test. A posteriori comparisons of variables among brood size categories were made using Tukey's multiple-range test.

Each brood was assigned a value for manipulated brood size (MBS) that matched the number of young by which its clutch size was changed (i.e., MBS = 0 if brood size = clutch size; MBS = +2 if two young added to brood). The individual-optimization hypothesis (IOH; Perrins and Moss 1975) assumes that clutch size reflects parental/territorial quality. Hence, pairs that began with a larger clutch should fledge more young than pairs that began with a smaller clutch when both are given the same number of young to raise (Nur 1986). Regardless of the number of eggs that are laid, however, all measures of success and nestling quality (i.e., mass and size) should decline when brood size exceeds clutch size. The IOH further predicts that females that laid different-sized clutches will produce young of equal quality if brood size equals clutch size (= control pair). But, when pairs that began with different-sized clutches are forced to raise broods of equal size (and all the young fledge), nestling quality will be higher in control than in enlarged broods (Nur 1986). I tested for these relationships using ANOVA and multiple regression. Details are provided in the Results.

The bad-years hypothesis (BYH) assumes that unpredictable (but regularly occurring) years of low success of large broods favors the evolution of clutches that are smaller than the brood size that in most years maximizes success (Murphy 1968, Boyce and Perrins 1987). I tested for the possible importance of the "bad-year" phenomenon for the evolution of the kingbird's clutch size by relating fledging success to brood size for each year individually. The BYH predicts that fledging success in good years will increase linearly with brood size and that the largest broods will be the most productive. Conversely, fledging success of large broods is predicted to be low during bad years, and pairs raising intermediate-sized broods will attain the greatest success. I used linear and polynomial regression to determine whether productivity increased linearly or quadratically with brood size. Evidence in support of the BYH would appear as a negative and significant coefficient for the quadratic term (= [[brood size].sup.2]) in the second-order polynomial regression of fledging success against brood size.

The hypothesis that reproduction is costly to adults makes several predictions. First, I tested for a negative impact of parental care on adult body condition by comparing adult body mass at the end of the nestling period to brood size. This analysis was complicated by the fact that mass varies among individuals because they differ in structural size. To control for the confounding effect of size, I used the first axis from a principal-components analysis (PCA) of wing chord, and tarsus and bill length to approximate body size. All three variables loaded positively on the first axis, which explained 53% of the total variation in size and shape. I then regressed body mass against axis one from the PCA and two other variables that varied with body mass: time of day and date of capture. The residuals from this multiple regression are thus corrected for differences among individuals in size, and time and date of capture. In the text below I refer to this corrected measure as both "mass" and "condition," but it i s best viewed as a measure of overall body condition.

The prediction that return rate should decline with increasing reproductive effort was tested by comparing adult return rates to brood size, and the combination of clutch size and manipulated brood size, using logistic regression (Pettifor 1993b). A difficulty in estimating survival from return data is that the probabilities of survival, resighting/recapture, and permanent dispersal from the area all contribute to observed return rates (Boulinier et al. 1997). Male kingbirds exhibit very high site fidelity (Murphy 1996a), and though females are more likely than males to disperse after nest failures, most females also return to former breeding sites (Murphy 1996a). Nonetheless, to reduce the probability of biasing my estimates of survival, I eliminated all nests that failed to fledge at least one nestling from the analysis of return. I also limited my analysis of adult survival to birds banded up to and including the 1996 season. All birds were thus given at least 2 yr to reappear before they were recorded as missing and presumed dead. Given these precautionary moves, the naturally high site fidelity of kingbirds, and my intensive searches for color-banded individuals, I am confident that most of the survivors were detected. Adult costs of reproduction might also manifest themselves as future fecundity costs (e.g., Gustafsson and Sutherland 1988). I therefore used ANOVA and multiple regression to compare timing of breeding, clutch size, hatching success, the incidence of starvation, and fledging success in year X + 1 to a pair's brood size and manipulated brood size in year X. These latter analyses were limited to pairs in which neither adult's effort was manipulated in year X + 1.

Statistical tests were made using Statistix (version 4.1; Analytical Software 1994). In all tests I based comparisons on the number of broods (not number of nestlings), and unless otherwise stated accepted that differences were statistically signficant if P [less than or equal to] 0.05. Detecting small but statistically significant differences can often be difficult, especially when sample sizes are small. This raises the possibility of committing Type II errors (acceptance of false null hypothesis). I therefore also conducted power analyses (Borenstein and Cohen 1988) to evaluate my ability to reject false null hypotheses in favor of a correct alternative. P values in ANOVAs and multiple regressions are based on Type III sums of squares, which evaluate the significance of each variable after controlling for the other variables in the models. All multiple coefficients of determination ([R.sup.2]) were adjusted for the number of independent variables.

RESULTS

The Lack clutch size

Clutch size ranged between two and four eggs, but most females laid three eggs (62.1%, n = 503 clutches; Fig. 1). Clutch size declines with laying date (Murphy 1986a), and as a result of an 1 1-d span in average date of clutch initiation over the 9-yr period of study ([F.sub.8,523] = 21.67, P [less than] 0.001), clutch size varied slightly among years ([F.sub.8,494] = 1.90, P = 0.058). Removing the effect of laying date eliminated all annual differences in clutch size ([F.sub.8,494] 1.52, P = 0.15).

Under the assumptions that food is limited and costs of reproduction are passed onto offspring, I predicted that both the proportion of young to fledge and nestling quality would decline with increasing brood size. In agreement, the proportion of young to starve increased with brood size (Kruskal-Wallis test, H = 10.22, P = 0.037) due to the higher probability of starvation in broods of three, four, and five young (8%, 8%, and 6%, respectively) compared to broods of one (0%) and two (2%) young. Losses to predators accounted for [tilde]23% of young regardless of brood size ([F.sub.4,377] = 0.14, P 0.962). Unless young were known to have starved, the latter analyses assumed that partial brood losses were the result of predation if the remaining young fledged in good condition. I also conducted analyses assuming that all partial losses were due to starvation. The proportion of young lost to predators was still independent of brood size (20%; F = 0.10, P = 0.980), but the proportion to starve no longer varied with bro od size (H = 5.431, P 0.24). Thus, regardless of whether partial losses were assigned to predation or starvation, both the probability that a nest was depredated, failed for other reasons, or fledged young (G = 5.289, df 8, P 0.71), and the proportion of young to fledge (Table 2: see Cost hypothesis), were independent of brood size. Productivity thus increased linearly with the number of young in the nest (Table 2; Fig. 1). When depredated nests were excluded from the analyses, the latter finding held for both manipulated ([r.sup.2] = 0.365, df = 280, P [less than] 0.001) and control broods (clutch size = brood size; [r.sup.2] = 0.149, df = 146, P [less than] 0.001).

Intergenerational costs: offspring quality and survival

As predicted by the cost hypothesis, offspring mass declined with brood size (by [tilde]1.2 g for each additional nestling; Fig. 2A). Young in broods of one and two birds did not differ from one another, but all remaining comparisons of nestling mass differed among brood sizes (Tukey's test, P [less than] 0.05). Nonetheless, brood size accounted for only [tilde]16% of the overall variation (Table 2: Cost hypothesis). Ninth primary length also declined significantly with brood size (Table 2). Young in broods of one and two birds had primaries that were 1.0-1.7 mm longer than offspring in broods of 3-5 young (Fig. 2B). There was no indication that either tarsus or bill length varied with the number of young in the nest (Table 2, Fig. 2C and D). Failure to establish significance for the latter two variables might be related to the lower power of both tests (0.33 and 0.18, respectively).

Young stay with their parents for up to 5 wk after fledging (personal observation), and maximum parental provisioning occurs [tilde]10 d after the young leave the nest (Morehouse and Brewer 1968). Within the sub-sample of 81 broods that were tracked into the post-fledging period, the number of young to leave the nest increased with brood size ([r.sup.2] = 0.599, df = 79, P [less than] 0.0001), but the number of young to die after leaving the nest during the post-fledging period also increased with brood size ([r.sup.2] = 0.086, df 79, P = 0.008). As a result, only 28.8% of the variation in the number of young surviving to 30 d was associated with brood size (Table 3; Fig. 3). Little of the remaining variability in the number of surviving fledglings appeared to be associated with the quadratic term ([brood size] [2]) in the polynomial regression of number alive against brood size (Table 3: Cost hypothesis; Fig. 3). Likewise, fledgling survival was independent of fledging date, nestling mass, and lengths of the nint h primary or tarsus (Table 3). However, due to the low power of all of the tests (Table 3) and the existence of significant correlations between brood size and both number of surviving young and nestling mass (P [less than or equal to] 0.002), I cannot conclude unequivocally that fledging mass had no impact on future survival. The only indication that survival might be related to size was a nonsignificant tendency for survival and tarsus length to be positively associated (Table 3). Bill length was not measured in 17 broods. When I limited my analyses to the 64 broods with complete data, all previous results remained unchanged. However, a significant and positive partial correlation existed between number of young alive at 30 d and bill length (Table 3). Brood size (P [less than] 0.001) and bill length (P = 0.045) together accounted for 30.0% of the variation in the number of surviving young. Overall, my results suggest that fledgling mortality did not occur disproportionately among larger broods during the p ost-fledging period, but that young with long bills (and possibly longer tarsi) tended to be more likely to survive.

Single offspring from 22 first nests returned as adults (10.2% of broods, or 3.3% of banded young). I compared fledging date and all measures of nestling size and condition between broods that did and did not produce recruits to evaluate whether any differences existed. Broods that yielded adults did not differ from other broods (n = 194) in either date of fledging (t = 0.67, P = 0.68), ninth primary length (t = 0.08, P = 0.94), tarsus length (t = 0.43, P = 0.67), bill length (t = 0.01, P = 1.00), or mass (t = 1.33, P = 0.18). The small number of known recruits resulted in low power (0.08-0.29) for all tests, and to attain significance the average differences between groups would have to have been 2-3 times greater than observed (and seven times greater for tarsus length). I note also that recruits were drawn from almost the entire range of fledging mass (Fig. 4), and that recruits tended to be lighter (32.2 g, 1 SD = 2.82) than young that were never seen again (32.8 g, 1 SD = 2.75, n = 194), contrary to the predictions of Lack's model. Logistic regression also indicated that the probability of being seen as an adult was not related to any combination of breeding date, mass, or other measure of nestling size (all P's [greater than] 0.50). Moreover, within-brood comparisons showed that the offspring to return was equally likely to have been above or below the brood's mean mass (Fisher exact test, P = 0.38), and primary (P = 0.18), bill (P = 0.33), or tarsus (P = 0.50) lengths. Finally, a comparison of the proportion of recruits to fledge from different-sized broods was virtually identical to the proportion of nestlings that fledged from each brood size (Fig. 5), a result that is consistent with the hypothesis that the per capita probability of recruitment was independent of brood size.

Individual optimization of reproduction

The individual-optimization hypothesis (IOH) predicts that parental ability varies positively with clutch size, but that enlargement of brood size is costly for all females. Thus, increasing brood size is expected to negatively influence nestling quality and the proportion of young to fledge--but given the same-sized brood, females that laid larger clutches should fledge more of the brood and the surviving young should be in better condition if all fledge (Nur 1986). Contrary to these predictions, I found that clutch size was negatively related to the proportion of young to fledge and with three of four measures of nestling size (Table 2: Individual-optimization hypothesis). The combination of clutch size and manipulated brood size also explained no more of the variation in any of the variables than did brood size (Table 2). These results suggest that all females were equally capable regardless of their original clutch size. For instance, comparisons of nestling mass to brood size showed that the decline in mass with increasing brood size was identical for females that laid clutches of either three or four eggs (Fig. 6).

The number of young to die in the 2-wk period after fledging increased significantly with the second-order polynomial of manipulated brood size ([[MBS].sup.2]); ([R.sup.2] = 0.124, df = 2.78, P = 0.006; MBS, P = 0.004 and [[MBS].sup.2], P = 0.009). Further analysis indicated that the number of young alive at 30 d was positively related to clutch size and [(MBS).sup.2] (Table 3: Optimization hypothesis). The three variables--clutch size, MBS, and [(MBS).sup.2]--accounted for 32% of the variation in the number of surviving young, and indicated that parents with enlarged broods tended to lose the extra young by -30 d of age. In fact, the number of young in enlarged broods that were alive at 30 d was identical to that of control broods when analyses were conducted separately for three- and four-egg-clutch nests (Fig. 7). The positive relationship with clutch size, which persisted after controlling for the effects of brood size, also suggested that parents that began with larger clutches had more young alive at 3 0 d, as predicted by the IOH. The brood's average fledging mass, and linear measures of nestling size did not explain additional variation in the number of surviving young (Table 3), but, again, low power and problems of multicollinearity among brood size, fledging mass, and number of fledged young make it difficult to draw firm conclusions. Restriction of the analysis to the 64 broods with information on bill length again produced a significant positive relationship between number of surviving young and bill length (Table 3).

Because of the small number of recruits, I combined all reduced broods (MBS = -2 and -1) into one category and all enlarged broods (MBS = +1 and +2) into another, and compared the proportion of young that fledged from reduced (0.182), control (0.591), and enlarged (0.227) broods to the proportion of recruits that fledged from reduced (0.223), control (0.567), and enlarged (0.210) broods. The virtually identical proportions (G = 0.226, df = 2, P = 0.89), suggest that recruitment was independent of manipulation.

Annual effects: bad years?

Analyses to this point have ignored possible annual differences in nesting success, but in fact the probability that a nest with hatchlings would fledge young varied greatly among years (Fig. 8). The "bad-years" hypothesis (BYH) posits that high annual variability in the fledging success of large broods favors the production of smaller clutches that have less variability in success (and thus higher probability of producing some young in all years). Because a rigorous test requires lifetime measures of success for genotypes laying different-sized clutches, I was only able to evaluate whether the observed pattern of annual variation in brood-size-dependent success was consistent with the evolution of small clutch size through risk avoidance. Predictions of the BYH are that fledging success will increase linearly with brood size in "good years" (Table 4: Linear model), but that fledging success will vary as a negative quadratic function of brood size in "bad years" (Table 4: Quadratic model). The latter would b e indicated by a negative and significant coefficient for the quadratic term ([[brood size].sup.2]). In eight of nine years productivity tended to increase linearly with brood size: the relationship was significant in six years and nearly significant in the other two years (Table 4). 1992 was the only year in which productivity showed no tendency to increase with brood size, and this was the result of a high nest-predation rate (46%) that was independent of brood size (G = 3.129, df = 3, P = 0.42). The quadratic term was negative in four of nine years (Table 4), but it was only once significant (1993). Ignoring 1992 (because of high nest predation) and based on a comparison of adjusted [R.sup.2] values, the linear model provided a better fit to the data in six of eight years, the quadratic model in one year, and in 1997 the linear and quadratic models performed equally well. My results suggest that environmental conditions were rarely poor enough to prevent broods of four and five young from fledging all of t heir young, and the largest broods were able to be raised in almost all years.

Observations from 1996, another year of low nesting success (Fig. 8), demonstrated convincingly that intrabrood losses of young due to the parents inability to feed them was often independent of brood size. Nearly a third (31.3%) of 147 nestlings starved in 1996 due mainly to a 36-h period of continuous and moderately heavy rain when most broods were being fed. Starvation never accounted for [greater than]4.5% of nestling deaths in the other eight years. Of the 5, 19, and 7 broods of 2, 3, and 4 young, respectively, that were active at the time of the rain, 25.8% experienced total failure by the end of the 36-h period. Broods with no nestling starvation were either younger (X = 6.1 d, 1 SD = 1.78, n = 7 broods) or older (X = 12.2 d, 1 SD = 1.14, n = 8) than broods with partial (X = 8.8 d, 1 SD = 2.28, n = 8) and complete (X = 9.4 d, 1 SD = 1.14, n = 8) starvation. Age differed significantly among the four groups (ANOVA: [F.sub.3,27] = 16.31, P [less than] 0.000), but average brood size varied only between 3. 0 and 3.1 ([F.sub.3,27] = 0.09, P = 0.96). The probability of nestling starvation was thus linked tightly with age rather than brood size.

Intragenerational costs of reproduction

Loss of body mass.--The increased feeding effort required of parents with large broods (Maigret and Murphy 1997) suggests that physiological stress increases with brood size, and the cost hypothesis predicts that parental body condition should decline as work load increases. The optimization model predicts the condition should decline with increased effort (i.e., enlargement of brood size), but that condition should also vary positively with clutch size. Predictions of the cost hypothesis were for the most part verified. Male body condition declined significantly with brood size (Fig. 9A) largely because males raising broods of two were 5.1% heavier than males tending broods of five (Tukey's test, P [less than] 0.05). Female body condition also tended to decline with increasing brood size, but the pattern was only marginally significant (Fig. 9A). Among pairs with broods of two and three young, males were in significantly better condition than females, but in larger broods male condition declined to the poin t that the parents were in equal condition (Fig. 9A). Patterns of variation in body condition only weakly conformed to the predictions of the optimization model. The decline in male body condition with increased manipulation was marginally significant (Fig. 9B). The latter relationship became significant (P = 0.02) when clutch size was included in a regression analysis, but, contrary to the predictions of the optimization model, condition declined with increasing clutch size (P = 0.036). Females that fed reduced broods tended to be in slightly better condition than females with control and enlarged broods, but the differences were not significant (Fig. 9B). The strength of the relationship between condition and manipulation increased (P = 0.07) when clutch size was entered in a multiple regression. However, the influence of clutch size was not significant (P = 0.56) and condition tended to vary inversely with clutch size.

Adult survival.--To evaluate the hypothesis that breeding patterns reflect trade-offs between current and future reproduction, I compared the probability of re turning to breed in year X + 1 to both brood size and manipulated brood size in year X for pairs of kingbirds that fledged young. For females, I also included year of breeding as a factor in the analysis since their return rates from 1992 (33.7%) and 1996 (47.3%) were significantly lower (Table 5) than rates in other years (mean = 62.1%, 95% confidence interval = 54.5- 69.7). For reference, in Fig. 10 I included the probability of return for birds that did not lay replacement clutches after their nest failed during incubation. Only 40% of the latter group of females returned the next year, no doubt because they dispersed to other sites (Murphy 1996a). Among successful breeders, brood size did not by itself explain variation in female return, but a second-order polynomial regression of brood size was significant (Table 5). Thus, females that raised bro ods of two and five young had lower probabilities of return than females that raised broods of three and four young (Fig. 10). Females that raised broods of two, three, and four young were 28.5%, 59.4% and 45.0%, respectively, more likely to return than were females that raised broods of five. Female return did not vary with the combination of clutch size and manipulated brood size, or with a second-order polynomial of manipulated brood size (Table 5).

Males exhibited different patterns. First, the probability of returning to breed declined steadily over the period of study (Table 6). Second, 70% of the males whose nests failed during incubation returned to breed the next year, and third, return rate appeared to be independent of success and overall productivity (Fig. 10). After controlling for the steady decline in return over the study period, return rate did not vary with any combination of brood size, clutch size, and/or manipulated brood size (Table 6).

Future fecundity.--Among females that successfully fledged young in year X and that were not manipulated in the following year, both breeding date ([F.sub.4,76] = 0.64, P 0.60) and clutch size ([F.sub.4,81] 0.70, P = 0.60) in year X + 1 were independent of brood size in year X. To facilitate comparisons of manipulated brood size, I combined all reduced broods into a single category (reduced) and enlarged broods into a second ( enlarged) and compared breeding statistics among the reduced, control, and enlarged categories. Neither date nor clutch size was affected by the previous year's manipulation of brood size (Table 7). The number of addled (i.e., unhatched) eggs showed a slight tendency to increase in year X + 1 with increased effort in the previous year (Table 7). Nonetheless, brood size, and the number of young to starve or fledge did not vary with the previous season's experimental modification of brood size (Table 7). It is worth noting, however, that females that raised reduced broods in year X were 58% more productive in year X + 1 than were females that raised enlarged broods. As with females, male timing of breeding and clutch size were unrelated to the previous year's reproductive activity (Table 7). Like-wise, the failure of eggs to hatch and other variables showed no tendency to decline with increasing effort in the past year.

Two potential problems exist with the analyses reported in Table 7. First, most losses of eggs and nestlings were due to predators. Nest predation is largely an extrinsic factor that is probably unrelated to the previous season's breeding effort, and this may mask residual effects from the earlier manipulation. Second, the data are a composite of nests, some of which were not followed completely through the nest cycle. A better analysis for detecting carryover effects between seasons would be to track nests for which a complete history is available and that escaped predation. Data were available for 45 females and 59 males. Among the females, breeding date ([F.sub.2,42] = 1.31, P = 0.28) and clutch size (Fig. 11) did not vary with manipulation. However, the number of addled eggs increased linearly with manipulation in the past year ([F.sub.2,42] = 5.03, P = 0.01), and as a consequence brood size tended to decline with effort in the previous season (Fig. 11). The number of young to starve was twice as high for females that raised an enlarged brood in the past year (0.38 young/brood) as compared to females that raised reduced broods (0.19 young/brood; control broods = 0.24 young/brood), but the difference was not significant ([F.sub.2,42] = 0.28, P = 0.76). Total within-nest losses (addled eggs + starved young) increased linearly and significantly with manipulated effort in the past season ([F.sub.2,42] = 3.46, P = 0.041), and the result was that the number of fledged young in year X + 1 declined (P = 0.06) with increasing female effo rt in the previous year (Fig. 11). Identical analyses were conducted with brood size in place of the reduced, control, and enlarged categories. In all cases, the same patterns emerged (i.e., high reproductive effort in year X was associated with reduced success in year X + 1), but brood size explained less of the variation than did manipulation (results not shown).

Among males, performance was largely independent of events in the previous season. Breeding date ([F.sub.2,42] = 0.63, P = 0.54), clutch size (Fig. 11), number of addled eggs ([F.sub.2,42] = 0.07, P = 0.93), and brood size (Fig. 11) were all independent of the previous season's manipulation. Nestling starvation was slightly more common in broods tended by males that had raised enlarged broods in the previous year ([F.sub.2,42] = 1.67, P = 0.20), but neither total within-nest losses ([F.sub.2,42] = 1.42, P = 0.25) nor number of fledglings (Fig. 11) varied with manipulation in the past season. Substituting brood size for the reduced-, control-, and enlarged-brood categories did not alter the results (results not shown).

DISCUSSION

My experiments indicate that Eastern Kingbird reproductive output was not limited by the parent's ability to find and deliver food during the nestling period. Broods of five, which have never been recorded in this population and which represent a 67% increase in the number of young above the modal clutch size, were the most productive. Large broods were no more likely to be depredated than were small broods. The only indication that adults had difficulty caring for enlarged broods was the steady decline in nestling mass with increased brood size, and the longer primaries of young raised in broods of one and two compared to larger broods. Although small size at fledging did not appear to have long-term negative effects on either the probability of surviving the fledgling period or recruitment, I cannot completely eliminate the possibility that survivorship was reduced when individuals were light or small at the point of fledging because (a) sample sizes (and thus statistical power) of most of my tests tended to be low and (b) analyses were confounded by the existence of multicollinearity between several predictor variables. Nonetheless, I again note that recruits tended to be lighter than individuals that were never seen again, a pattern that lends no support to the hypothesis that small fledglings were disadvantaged. The significant positive relationship between bill length and number of surviving young (Table 3) is difficult to interpret. However, given that bill length did not vary with brood size, I suggest that differences in bill length reflect heritable variation (e.g., Schluter and Smith 1986, Merila and Gustafsson 1993) and parental quality, rather than brood-size-dependent parental care. Taken as a whole, it appears that parents did not pass the majority of the cost of raising extra young along to the brood (Mauck and Grubb 1995, Wernham and Bryant 1998), which leads to a firm rejection of Lack's basic model, and no support for the intergenerational cost hypothesis.

The relative ease with which kingbirds raised broods of five, and the absence of severe survival cost to young that fledged from enlarged broods, also allow me to eliminate several other hypotheses that have been proposed as explanations for why birds with altricial young often fail to lay more eggs. Mountford's (1968) "cliffedge" hypothesis can be rejected because I found no evidence for its critical prediction of asymmetrically low survivorship of young from large broods, regardless of whether survival was measured at fledging, 12- 14 d after fledging, or to the point of recruitment. The parasitism-insurance hypothesis (PIH; Power et al. 1989) can also be rejected based on patterns observed in this and a companion work. The PIH was a tenable hypothesis during the early phase of my study because evidence showed that (a) nestlings in broods of five often starved (Murphy 1983b), (b) kingbirds accept foreign eggs from conspecifics (Bischoff and Murphy 1993), and (c) intraspecific brood parasitism occurs (McKit rick 1990). However, the present and much larger data set shows that broods of five are raised as effectively as smaller broods. Moreover, direct observations of laying patterns and DNA fingerprinting have yielded no evidence of brood parasitism (D. L. Rowe, M. T. Murphy, R. C. Fleischer, and P. G. Wolf, unpublished manuscript). Thus, two essential conditions for the evolution of small clutch size as a response to frequent intraspecific brood parasitism do not exist in kingbirds.

The hypothesis that stochastic variation in the conditions for rearing young favors low reproductive effort has received much support from theoreticians (e.g., Murphy 1968, Gillespie 1977, Frank and Slatkin 1990). Avian ecologists have given it little attention probably because adequate tests require many years of data to detect the small advantages that would favor genotypes following a bet-hedging strategy. The essential argument is that selection against high variance in success favors smaller clutches because large broods fail regularly, but unpredictably, as conditions vary among years (Boyce and Perrins 1987). Although Boyce and Perrins (1987) initially suggested that high variance in the success of large Great Tit (Parus major) broods favored smaller clutches, Liou et al.'s (1993) reanalysis of their data showed that the putative advantage of following a bet-hedging strategy was a statistical artifact of the small sample of large broods. Results of earlier brood-size manipulation experiments that I co nducted on kingbirds (Murphy 1983b) suggested that large broods were far more likely to have young starve during cool and rainy weather because the primary food supply, flying insects, disappears. Given that poor weather can occur at any time during the breeding season, I proposed that large broods probably failed frequently enough to favor females that produced smaller clutches. However, my current data set provides no support for the hypothesis that highly variable conditions for feeding young is the basis for the current small clutch size of kingbirds. The probability of starvation did not differ among broods of three, four, and five young, and overall, nestling starvation never claimed more than 4.5% of young in any year except for 1996 (31%). But even during that very poor year the probability of starvation (either partial or complete brood loss) was independent of brood size. As a consequence, in all but two of the nine years of this study the largest broods tended to be the most productive. I thus have no evidence that selection for low variance in nesting success arising from the frequent starvation of young in large broods has favored the evolution of small clutch size in kingbirds.

Individual optimization

The underlying premise of the individual-optimization hypothesis (IOH) is that females produce large clutches only when they are capable of raising large broods. Lack (1966; see also Perrins and Moss 1975) offered this as an explanation for why clutch sizes above the population's mode were often the most productive (see reviews by Stearns [1992] and VanderWerf [1992]). The differential capacity for raising young may be a consequence of variation in territory quality (Hogsedt 1980) or intrinsic differences among females (Goodburn 1991). To date, the IOH has been tested in relatively few species, but supporting evidence has been found in mapies (Pica pica; Hogstedt 1980), Blue Tits (Parus caeruleus; Pettifor 1993a [but see Nur 1986]), Collared Flycatchers (Ficedula albicollis; Gustafsson and Sutherland 1988); and Great Tits (Pettifor et al. 1988, Tinbergen and Daan 1990). Yellow-headed Blackbirds (Xanthocephalus xanthocephalus) do not appear to individually optimize reproduction (Barber and Evans 1995), but co nclusions from this latter study must be viewed cautiously since young were followed only to the point of fledging.

My current work suggests that kingbirds also individually optimized their effort. I find it noteworthy that if I had ended my observations at the point of fledging I would have concluded that kingbirds did not individually optimize their effort because (1) enlarged broods produced the most fledglings (Fig. 1); (2) regardless of whether three or four eggs had been laid, the same number of young fledged from broods of equal size; and (3) although nestling body mass declined as brood size increased, mass did not differ between nests that began with different numbers of eggs but that held the same number of young (Fig. 6). Thus, up to the point of fledging it did not appear that any differences existed in the parental abilities of pairs tending nests that began with three or four eggs. However, by day 12 of the post-fledging period more offspring died in broods that had been enlarged. Indeed, broods that had been enlarged by one nestling produced no more surviving offspring than control broods that began with th e same number of eggs (Fig. 7). Morehouse and Brewer's (1968) observation that parental feeding rate of the young peaked [tilde]10 d after fledging possibly ex plains why young were lost mainly after, rather than before, leaving the nest. That young from enlarged broods nonetheless entered the breeding population suggests that survival through the 2-wk period after fledging was critically important, and that parental ability was tested most severely at this stage of the reproductive cycle (see also Sullivan [1989] and Magrath [1991] for similar conclusions).

Intragenerational costs of reproduction

Although the evidence presented thus far suggests that kingbird reproductive effort is to at least some extent individually optimized, the IOH fails to explain why kingbirds in this population never lay five-egg clutches. The high feeding rates (Morehouse and Brewer 1968) and loss of extra young during the post-fledging period (this study) are consistent with an assumption of food limitation and individual optimization, but it is not clear that the extra young starved. For instance, I found no relationship between the average mass or structural size of nestlings just before fledging and their survival through either the post-fledging period or to the point of recruitment. Even within single broods, size did not predict survivorship. An equally valid interpretation for the loss of young in large broods is that they were more prone to predation because the parents could not guard them effectively (Safriel 1975). But the answer, I suspect, to why kingbirds in this population never produce five-egg clutches is t hat broods of five severely reduce a female's future prospects of reproduction.

Earlier reviews of experimental investigations in birds of the impact of current breeding effort on adult survival and fecundity in the next year (Linden and Moller 1989, Dijkstra et al. 1990, Nur 1990, Stearns 1992) have detected a negative impact of reproduction on adult survival in [tilde]40% of studies. Although sample sizes were smaller, future fecundity costs tended to be slightly more common. Additional work since then breaks down in similar proportions. No costs of reproduction were reported by Tinbergen and Daan (1990), Torok and Toth (1990), Wiggins (1990), Wheelwright et al. (1991), Pettifor (1993b), or Orell et al. (1996), whereas costs were demonstrated by Jacobsen et al. (1995; but see Daan et al. (1996), Boulinier et al. 1997). Golet et al. (1998), and Wernham and Bryant (1998). Among passerines, I have been able to find only 14 experimental studies in which costs of reproduction (primarily adult survival) were measured between years (Table 8; nine species, data up through 1997). All of the studies listed in Table 8 used brood-size manipulations to vary costs, but most were based on small sample sizes, and as a consequence the power of most tests was low. Low power is likely to lead to the acceptance of false null hypotheses of no costs associated with reproduction when costs may have in fact existed. In an attempt to compensate for the low power as I evaluated the results, I categorized all tests that were in the direction predicted by the cost hypothesis as significant if P [less than] 0.50. Using these liberal criteria, 7 of 12 (58%) comparisons (sex and species combinations) of adult body mass and 9 of 28 (32%) comparisons of adult survivorship yielded evidence of increased costs when adults were forced to raise enlarged broods. Fewer data were available for future reproduction and productivity, but the results tend to suggest that increased effort had little impact on future breeding. Brood-size enlargement was followed by a delay in breeding in the next year in 2 of 9 cases (22%), by smaller clutc h size in 1 of 13 comparisons (8%), and by reduced productivity in 3 of 13 comparisons (23%).

Two major points can be made from Table 8. Increases in brood size were frequently followed by loss of body mass (for results in other studies of passerines see Bryant [1988], Johnston [1993], Hillstrom [1995]; but see Sanz and Moreno [1995]). On the other hand, the majority of studies (68%) suggest that enlargement of brood size was not followed by increased adult mortality, despite the liberal criteria used as evidence for costs. Future breeding events were affected by past reproductive activity even less frequently. For instance, a female's future clutch size was reduced only one of eight times (12.5%) after being forced to raised enlarged broods (Table 8). Although the power of many of these tests was low, they provide no compelling evidence for the existence of reproductive costs among passerines. However, this conclusion must be viewed in light of the fact that the sample of species is very biased towards secondary cavity nesters (7 of 8 species) and the genus Parus (7 of 15 studies). This leads to my second point, which is that essentially nothing is known about costs of reproduction in open-cup-nesting passerines because experimental work has been conducted on only two species--rooks (Corvus frugileus) and kingbirds. Given the differences in clutch size and survival rates between these ecological groups (see Martin 1995), there is no reason to believe that conclusions about secondary-cavity-nesting species apply to open-cup nesters. Indeed, the high reproductive rates and low survival rates that are typical of secondary-cavity nesters (Martin 1995) probably leave little room for the detection of additional costs associated with enlargement of brood size, and we should perhaps not find it surprising that so few studies of passerines have found evidence of reproductive costs. The most appropriate conclusion that can be made regarding our current state of knowledge is that elevated reproductive effort does not significantly increase reproductive costs in most species of secondary-cavity-nesting birds, and t hat reductions in effort (e.g., Golet et al. 1998, Wernham and Bryant 1998) are more likely to detect the existence of reproductive costs in this group. Conclusions for other ecological groups await the collection of basic data.

Results of brood-size enlargements for Eastern Kingbirds suggest that reproduction was costly, at least for adult females. Previous work has shown that parents steadily increase feeding rates between broods of one and three young but that broods of five appear to be near the parent's maximum feeding rate (Fig. 2 in Maigret and Murphy 1997). I found that body condition of both sexes tended to decline with increasing brood size, and in both sexes parents caring for five young were in the poorest condition (Fig. 9A). Roughly 62% of females that raised broods within the normal range for this population (2-4 young) returned to breed in the following year, but only 42% of females returned when given broods of five. Given the apparent ceiling on feeding rate and poorer body condition of females feeding five young, the low return rate for these females seems likely to reflect true mortality. The alternative explanation, which seems highly unlikely, is that females that successfully raised broods of five dispersed to new breeding sites. The low return rate of females that raised broods of only one or two young is best explained by dispersal after low reproductive success (Murphy 1996a). Indeed, Nur (1988) found that female Blue Tits given the smallest broods were the most likely to disperse in the following year. The low return rates for females that bred in 1992 and 1996, the years of low nesting success (Fig. 8), support this position.

Elevated reproductive effort in the past year also carried substantial fecundity costs for surviving females. Although breeding date and clutch size were unaffected, hatching failure and total within-brood losses (hatching failure + starved young) were significantly higher among females that raised enlarged broods in the previous year. The result was that fledging success in year X + 1 declined linearly with reproductive effort in year X (Fig. 11). That the experimental manipulation proved to be better than brood size at explaining variation in future reproductive performance adds support to my previous claim that females also individually optimized their clutch size.

Patterns among males were different. Body condition declined significantly with brood size, but breeding appeared to carry no survival cost (Fig. 11, Table 6). Males that did not feed young (because nests failed during incubation) and males that raised 1-5 young had identical survival rates. This seems at odds with my finding that male condition declined with increasing brood size, but I suspect that the explanation is that males have much lower feeding rates than females at all brood sizes (Woodard and Murphy 1999) and therefore males remain in overall better condition. Interestingly, once brood size exceeded the population's modal size (three young) males and females were in equal condition. Future male productivity was also unaffected by past effort, but this is not surprising since the major future fecundity cost appeared to be increases in the number of addled eggs.

To date the strongest evidence for the existence of reproductive costs in passerines comes from studies of aerial-foraging birds (Table 8). Among flycatchers belonging to two families, enlargement of brood size in Collared (Ficedula albicola) and Pied (F. hypoluca) Flycatchers and Eastern Kingbirds has shown statistically significant negative impacts on adult body mass (three of three species), survival (two of three species) and future productivity (two of two species). In a non-experimental study, Bryant (1979) also demonstrated significant reproductive costs in House Martins (Delichon urbica): double-brood females survived at only half the rate of single-brooded females. Possible mechanisms mediating the trade-off between reproduction and survival include parasite infestation or interruption of feather replacement prior to migration. For instance, Moller (1993) showed that swallows (Hirundo rustica) with ectoparasites produced fewer young over the season than did controls, but that the effects were most p ronounced among birds that reared enlarged broods. In addition, Siikamaki et al. (1997) found increases in haematozoan infections in Pied Flycatchers that were forced to feed additional young. Female Pied Flycatchers that raised large or enlarged broods also delayed their post-breeding moult in comparison to females that raised fewer young (Siikamaki et al. 1994).

Although flycatchers and swallows forage differently (Hespenheide 1971), food availability (Bryant 1975) and foraging success (Davies 1977, Murphy 1987) decline sharply in both groups during cool and rainy weather. Moreover, daily energy expenditure tends to be high in aerial foragers because energy use is positively linked to flight time (Westerterp and Bryant 1984, Bryant 1988). Aerial foragers may exhibit costs more frequently than other species because of this combination of high energy use and high daily variability in insect abundance. Frequent energetic stress, especially when raising large broods, may compromise immune systems or delay moult. Tree Swallows (Tachycineta bicolor) appear to be an exception. The lack of evidence of costs in Tree Swallows (Table 8) may be due to the fact that they lay one of the largest clutches within the Hirundinidae (Ramstack et al. 1998), which leaves little room to detect costs when they are forced to raised enlarged broods.

Synthesis

Kingbirds in this and most other populations (Murphy 1983a, 1986; but see Blancher and Robertson 1985) rarely lay five eggs despite an ability to raise artificially enlarged broods of five young. I have tested six hypotheses for why this pattern exists, and the results allow me to reject both Lack's (1947, 1966) basic and extended (intergenerational-cost) model of food limitation. The cliff-edge (Mountford 1968), parasitism-insurance (Power et al. 1989), and bad-years (Boyce and Perrins 1987) hypotheses also fail to provide an answer for why kingbirds do not lay larger clutches. I have not been able to reject predictions of either the individual-optimization or cost hypotheses, and I propose that kingbird clutch sizes are influenced by both phenomena.

Ultimately, clutch size is probably limited by costs of reproduction. Kingbirds are relatively long lived and theory predicts that parents should not jeopardize future reproductive potential for short-term gains when there is a high probability of surviving to breed again (Bulmer 1985, Stearns 1992; see Weimerskirch 1992, Werhham and Bryant 1998). Based on empirically derived survival rates, average life expectancy can be calculated as (2 -- M)/2M where M is equal to 1 - annual survival (Gill 1990). Assuming a constant probability of mortality for females that raised different-sized broods, predicted life expectancies for females in this population that raised three (M = 0.32), 4 (M = 0.39), or five (M = 0.58) young were 2.6, 2.1, and 1.2 yr, respectively. Thus, raising a brood of five severely reduced a female's potential lifetime productivity by limiting her to (in all probability) only a single breeding season. Despite the fact that survival was nearly the same for females that raised three or four young, and that both were likely to survive through two breeding seasons, clutches of three predominated. The reason why females did not lay more four-egg clutches may be that reproductive costs are cumulative (Gustafsson and Part 1990, McCleery et al. 1996; but see Wheelwright et al. 1991), and broods of three in the long run may produce the least stress and highest lifetime production of young. Indeed, surviving females that raised enlarged broods produced fewer young in their first breeding attempt after returning. The decision to lay three or four eggs is probably determined by a female's parental ability, but whether individual differences in ability reflect intrinsic properties of females or their territories is unknown.

I did not directly test predictions of the nest-predation hypothesis, and therefore cannot exclude it as a possible explanation for why kingbirds do not lay larger clutches. Indeed, given that nest predators destroy nearly 50% of nests every year (and sometimes more, Fig. 7), it would be rash to claim that threats of nest predation have not influenced the evolution of clutch size in kingbirds. Kingbird parents spend over a quarter of each hour in nest defense behavior (Woodard and Murphy 1999), and all other parental duties come at the expense of nest watching (Rosa and Murphy 1994, Woodard and Murphy 1999). There is good reason to believe that nest predation, as it interacts with food supply (Martin 1992) and costs of reproduction (Julliard et al. 1997), has contributed to the evolution of the normal three-egg clutch of kingbirds. Much work remains before the question is settled, but in the end it seems likely that single-factor explanations for why kingbirds and other species reproduce at a given rate will prove unsatisfactory.

ACKNOWLEDGMENTS

Over the years I have been assisted by many students that have served as both friends and colleagues. In particular I would like to thank the following students for their valuable contributions to my research and the data in this paper in particular: Brian Armbrecht, Cathy Bischoff, Judy Cole, Kerri Cornell, Charity Cummings, Brad Davey, Maria Gallagher, Jennifer Maigrett, Shawn Martin, Michael Miller, Michael Palmer, Aaron Pierce, Joy Ramstack, Stephanie Rosa, Diane Rowe, Andrea VanBuskirk, Ekaterini Vlamis, Jason Woodard, and Staci Wunder. Financial support for my work was provided by NSF grant BSR 9106854, the DANA Foundation, and numerous Trustees' Grants from Hartwick College. I am also grateful to the many landowners that gave me permission to study the birds that nested on their property, and to friends (Douglas A. Hamilton, Stanley K. Sessions, Robert Titus) who have encouraged and supported my work. L. Scott Forbes and an anonymous reviewer pruvided considerable constructive criticism to an earlier version of the manuscript. Most importantly, I thank my family for the numerous years of tolerance and understanding.

(1.) Department of Biology, Hartwick College, Oneonta, New York 13820 USA. E-mail: murphym@hartwick.edu

LITERATURE CITED

Analytical Software. 1994. Statistix. Version 4.1. Analytical Software. Tallahassee, Florida, USA.

Askenmo, C. 1977. Effects of addition and removal of nestlings on nestling weight, nestling survival, and female weight loss in the Pied Flycatcher Ficedula hypoleuca (Pallus). Ornis Scandinavica 8:1-8.

Askenmo, C. 1979. Reproductive effort and return rate of male Pied Flycatchers. American Naturalist 114:748-753.

Barber, C. A., and R. E. Evans. 1995. Clutch-size manipulation in the Yellow-headed Blackbird: a test of the individual optimization hypothesis. Condor 97:352-360.

Bischoff, C. M., and M. T. Murphy. 1993. The detection of and responses to experimental intraspecific brood parasitism in Eastern kingbirds. Animal Behaviour 45:631-638.

Blancher, P. J., and R. J. Robertson. 1985. Site consistency in kingbird breeding performance: implications for site fidelity. Journal of Animal Ecology 54:1017-1027.

Borenstein, M., and J. Cohen. 1988. Statistical power: a computer program. Lawrence Earlbaum Associates, Hillsdale, New Jersey, USA.

Boulinier, T., G. Sorci, J. Clobert, and E. Danchin. 1997. An experimental study of the costs of reproduction in the kittiwake Rissa tridactyla: comment. Ecology 78:1284-1287.

Boyce, M. S., and C. M. Perins. 1987. Optimizing Great Tit clutch size in a fluctuating environment. Ecology 68: 142-153.

Bryant, D. M. 1975. Breeding biology of house martins Delichon urbica (L.) in relation to aerial insect abundance. Ibis 117:180-216.

Bryant, D. M. 1979. Reproductive costs in the House Martin (Delichon urbica). Journal of Animal Ecology 48:655-675.

Bryant, D. M. 1988. Energy expenditure and body mass changes as measures of reproductive costs in birds. Functional Ecology 2:23-34.

Bulmer, M. G. 1985. Selection for iteroparity in a variable environment. American Naturalist 126:63-71.

Charnov, E. L., and J. R. Krebs. 1974. On clutch size and fitness, Ibis 116:217-219.

Cooch, B. G., and R. E. Ricklefs. 1994. Do variable environments significantly influence optimal reproductive effort in birds? Oikos 69:447-459.

Daan, S., C. Deerenberg, and C. Dijkstra. 1996. Increased daily work precipitates natural death in the Kestrel. Journal of Animal Ecology 65:539-544.

Davies, N. B. 1977. Prey selection and the search strategy of the Spotted Flycatcher (Muscicapa striata): a field study of optimal foraging. Animal Behaviour 25:1016-1033.

Davis, D. E. 1955. Observations on the breeding biology of kingbirds. Condor 57:208-212.

DeSteven, D. 1980. Clutch size, breeding success and parental survival in the Tree Swallow (Iridoprocne bicolor). Evolution 34:278-291.

Dijkstra, C., A. Bult, S. Bijlsma, S. Daan, T. Meijer, and M. Zijlstra. 1990. Brood size manipulation in the Kestrel (Falco tinunculus): effects on offspring and parent survival. Journal of Animal Ecology 59:269-285.

Frank, S. A., and M. Slatkin. 1990. Evolution in a variable environment. American Naturalist 136:244--260.

Gill, F. B. 1990. Ornithology. W. H. Freeman and Company, New York, New York, USA.

Gillespie, J. H. 1977. Natural selection for variances in offspring numbers: a new evolutionary principle. American Naturalist 111:1010-1014.

Godfray, H. C. 1987. The evolution of clutch size in parasitic wasps. American Naturalist 129:221--233.

Godfray, H. C., L. Partridge, and P. H. Harvey. 1991. Clutch size. Annual Review of Ecology and Systematics 22:409--429.

Golet, G. H., D. B. Irons, and J. A. Estes. 1998. Survival costs of chick rearing in Black-legged Kittiwakes. Journal of Animal Ecology 67:827-841.

Goodburn, S. F. 1991. Territory quality or bird quality? Factors determining breeding success in the magpie Pica pica. Ibis 133:85--90.

Gustafsson, L., and T. Part. 1990. Acceleration of senesence in the Collared Flycatcher Ficedula hypoleuca by reproductive costs. Nature 347:279--281.

Gustafsson, L., and W. J. Sutherland. 1988. The costs of reproduction in the Collared Flycatcher Ficedula albicollis. Nature 335:813--815.

Hegner, R. E., and J. C. Wingfield. 1987. Effects of brood-size manipulations on parental investment, breeding success, and reproductive endocrinology of House Sparrows. Auk 104:470--480.

Hespenheide, H. A. 1971. Food preferences and the extent of overlap in some insectivorous birds, with special reference to the Tyrannidae. Ibis 113:59--72.

Hillstrom, L. 1995. Body mass reduction during reproduction in the Pied Flycatcher Ficedula hypoleuca: physiological stress or adaptation for lowered costs of locomotion? Functional Ecology 9:807--817.

Hochachka, W., and J. N. M. Smith. 1991. Determinants and consequences of nestling condition in Song Sparrows Journal of Animal Ecology 60:995--1008.

Hogstedt, G. 1980. Evolution of clutch size in birds: adaptive variation in relation to territory quality. Science 210:1148--1150.

Jacobsen, K.-O., K. E. Erikstad, and B.-E. Saether. 1995. An experimental study of the costs of reproduction in the kittiwake Rissa tridactyla. Ecology 76:1636--1642.

Johnston, R. D. 1993. The effect of direct supplementary feeding of nestlings on weight loss in female Great Tits Parus major. Ibis 135:311--314.

Julliard, R., R. H. McCleery, J. Clobert, and C. M. Perrins. 1997. Phenotypic adjustment of clutch size due to nest predation in the Great Tit. Ecology 78:394--404.

Lack, D. 1947. The significance of clutch size. Ibis 89:302--352.

Lack, D. 1966. Population studies of birds. Clarendon, Oxford, UK.

Law, R. 1979. Optimal life histories under age-specific predation. American Naturalist 114:399--417.

Lima, S. L. 1987. Clutch size in birds: a predation perspective. Ecology 68:1062--1070.

Linden, M., and A. P. Moller. 1989. Cost of reproduction and variation of life history traits in birds. Trends in Ecology and Evolution 4:367--371.

Liou, L. W., T. Price, M. S. Boyce, and C. M. Perrins. 1993. Fluctuating environments and clutch size evolution in Great Tits. American Naturalist 141:507--516.

Lloyd, D. 1987. Selection of offspring size at independence and other size versus number strategies. American Naturalist 129:800--817.

Magrath, R. D. 1991. Nestling weight and juvenile survival in the blackbird, Turdus merula. Journal of Animal Ecology 60:335--351.

Maigret, J. L., and M. T. Murphy. 1997. Costs and benefits of parental care in Eastern Kingbirds. Behavioral Ecology 8:250--259.

Martin, T. E. 1992. Interaction of nest predation and food limitation in reproductive strategies. Current Ornithology 9:163--197.

Martin, T. E. 1995. Avian life history evolution in relation to nest sites, nest predation, and food. Ecological Monographs 65:101--127.

Mauck, R. A., and T. C. Grubb, Jr. 1995. Petrel parents shunt all experimentally increased reproductive costs to their offspring. Animal Behaviour 49:999--1008.

McCleery, R. H., J. Clobert, R. Julliard, and C. M. Perrins. 1996. Nest predation and delayed costs of reproduction in the Great Tit. Journal of Animal Ecology 65:96--104.

McKitrick, M. C. 1990. Genetic evidence for multiple parentage in Eastern Kingbirds (Tyrannus tyrannus). Behavioral Ecology and Sociobiology 26:149--155.

Meijer T., S. Daan, and M. Hall. 1990. Family planning in the Kestrel (Falco tinnunculus): the proximate control of covariation in laying date and clutch size. Behaviour 114:117--136.

Merila, J., and L. Gustafsson. 1993. Inheritance of size and shape in a natural population of Collared Flycatchers, Ficedula albicollis. Journal of Evolutionary Biology 6:375--395.

Moller, A. P. 1993. Ectoparasites increase the cost of reproduction in their hosts. Journal of Animal Ecology 62:309--322.

Monaghan, P., and R. G. Nager. 1997. Why don't birds lay more eggs. Trends in Ecology and Evolution 12:270--274.

Morehouse, E. L., and R. Brewer. 1968. Feeding of nestling and fledgling Eastern Kingbirds. Auk 85:44--54.

Morris, D. W. 1992. Optimum brood size: tests of alternative hypotheses. Evolution 46:1848--1861.

Morris, D. W. 1996. State-dependent life histories, Mountford's hypothesis, and the evolution of brood size. Journal of Animal Ecology 65:43--51.

Mountford, M. D. 1968. The significance of litter-size. Journal of Animal Ecology 37:363--367.

Murphy, G. I. 1968. Patterns in life history and environment. American Naturalist 102:390--404.

Murphy, M. T. 1983a. Ecological aspects of the reproductive biology of Eastern Kingbirds: geographic comparisons. Ecology 64:914--928.

Murphy, M. T. 1983b. Clutch size in the Eastern Kingbird: factors affecting nestling survival. Auk 100:326--334.

Murphy, M. T. 1986. Temporal components of reproductive variability in Eastern Kingbirds (Tyrannus tyrannus). Ecology 67:1483--1492.

Murphy, M. T. 1987. The impact of weather on kingbird foraging behavior. Condor 89:721--730.

Murphy, M. T. 1996a. Survivorship, breeding dispersal and mate fidelity in Eastern Kingbirds. Condor 98:83--93.

Murphy, M. T. 1996b. Eastern Kingbird (Tyrannus tyrannus). Pages 1-24 in A. Poole and F Gill, editors. The birds of North America. Number 253. The Academy of Natural Sciences, Philadelphia, Pennsylvania, USA.

Murphy, M. T., C. L. Cummings, and M. S. Palmer. 1997. A comparative analysis of habitat selection, nest site and nest success by Cedar Waxwings and Eastern Kingbirds. American Midland Naturalist 138:344--356.

Nager, R. G., C. Ruegger, and A. J. Van Noordwijk. 1997. Nutrient or energy limitation on egg formation: a feeding experiment in Great Tits. Journal of Animal Ecology 66:495--507.

Nur, N. 1984. The consequences of brood size for breeding Blue Tits. I. Adult survival, weight change and the cost of reproduction. Journal of Animal Ecology 53:479--496.

Nur, N. 1986. Is clutch size variation in the Blue Tit (Parus caeruleus) adaptive? An experimental study. Journal of Animal Ecology 55:983-999.

Nur, N. 1988. The consequences of brood size for breeding Blue Tits. III. Measuring the cost of reproduction: survival, future fecundity, and differential dispersal. Evolution 42: 351-362.

Nur, N. 1990. The cost of reproduction in birds: evaluating the evidence from manipulative and non-manipulative studies. Pages. 281-296 in J. Blondel, A. Gosler, J.-D. Lebreton, and R. McCleery, editors. Population biology of passerine birds. Springer-Verlag, Berlin, Germany.

Orell, M., and K. Koivula. 1988. Cost of reproduction: parental survival and production of recruits in the Willow Tit Parus montanus. Oecologia 77:423-432.

Orell, M., S. Rytkonen, K. Koivula, M. Ronkainen, and M. Rahlala. 1996. Brood size manipulations within the natural range did not reveal intragenerational cost of reproduction in the Willow Tit Parus montanus. Ibis 138:630-637.

Perrins, C. M. 1996. Eggs, egg formation and the timing of breeding. Ibis 138:2-15.

Perrins, C. M., and D. Moss. 1975. Reproductive rates in the Great Tit. Journal of Animal Ecology 44:695-706.

Pettifor, R. A. 1993a. Brood-manipulation experiments. I. The number of offspring surviving per nest in Blue Tits (Paras caeruleus). Journal of Animal Ecology 62:131-144.

Pettifor, R. A. 1993b. Brood-manipulation experiments. II. A cost of reproduction in Blue Tits (Parus caeruleus). Journal of Animal Ecology 62:145-159.

Pettifor, R. A., C. M. Perrins, and R. H. McCleery. 1988. Individual optimization of clutch size in Great Tits. Nature 336:160-162.

Platt, J. R. 1964. Strong inference. Science 146:347-353.

Power, H. W., E. D. Kennedy, L. C. Romagnano, M. P. Lombardo, A. S. Hoffenberg, P. C. Stouffer, and T. R. McQuire. 1989. The parasitism insurance hypothesis: why starlings leave space for parasitic eggs. Condor 91:753-765.

Ramsay, S. L., and D. C. Houston. 1997. Nutritional constraints on egg production in the Blue Tit: a supplementary feeding study. Journal of Animal Ecology 66:649-657.

Ramstack, J. M., M. T. Murphy, and M. R. Palmer. 1998. Comparative reproductive biology of three species of swallows in a common environment. Wilson Bulletin 110:233-243.

Reznick, D. 1985. Costs of reproduction: an evaluation of the empirical evidence. Oikos 44:257-267.

Reznick, D. 1992. Measuring the costs of reproduction. Trends in Ecology and Evolution 7:42-45.

Reznick, D., H. Bryga, and J. A. Endler. 1990. Experimentally induced life-history evolution in a natural population. Nature 346:357-359.

Risch, T. S., F. S. Dobson, and J. O. Murie. 1996. Is mean litter size the most productive? A test in Columbian ground squirrels. Ecology 76:1643-1654.

Rosa, S. M., and M. T. Murphy. 1994. Trade-offs and constraints on Eastern Kingbird parental care. Wilson Bulletin 106:668-678.

Roskaft, E. 1985. The effect of enlarged brood size on the future reproductive potential of the Rook. Journal of Animal Ecology 54:255-260.

Safriel, U. N. 1975. On the significance of clutch size in nidifugous birds. Ecology 56:703-708.

Sanz, J. J., and J. Moreno. 1995. Mass loss in brooding female Pied Flycatchers Ficedula hypoleuca: no evidence for reproductive stress. Journal of Avian Biology 26:313-320.

Schluter, D., and J. N. M. Smith. 1986. Genetic and phenotypic correlations in a natural population of Song Sparrows. Biological Journal of the Linnean Society 29:23-36.

Seger, J., and H. J. Brockmann. 1987. What is bet-hedging? Oxford Surveys in Evolutionary Biology 5:182-211.

Siikamaki, P., M. Hovi, and O. Ratti. 1994. A trade-off between current reproduction and moult in the Pied Flycatcher--an experiment. Functional Ecology 8:587-593.

Siikamaki, P., O. Ratti, M. Hovi, and G. F. Bennett. 1997. Association between haematozoan infections and reproduction in the Pied Flycatcher. Functional Ecology 11:176-183.

Skutch, A. F. 1949. Do tropical birds rear as many young as they can nourish? Ibis 91:430-455.

Slagsvold, T. 1982. Clutch size variation in passerine birds: the nest predation hypothesis. Oecologia 54:159-169.

Stearns, S. C. 1992. The evolution of life histories. Oxford University Press, Oxford, UK.

Sullivan, K. A. 1989. Predation and starvation: age-specific mortality in juvenile juncos (Junco phaenotus). Journal of Animal Ecology 58:275-286.

Tinbergen, J. M., and M. C. Boerlijst. 1990. Nestling weight and survival in individual Great Tits (Parus major). Journal of Animal Ecology 59:1113-1127.

Tinbergen, J. M., and S. Daan. 1990. Family planning in the Great Tit (Parus major): optimal clutch size as integration of parent and offspring fitness. Behaviour 114:161-190.

Torok, J., and L. Toth. 1990. Costs and benefits of reproduction of the Collared Flycatcher, Ficedula albicollis. Pages 307-319 in J. Blondel, A. Gosler, J.-D. Lebreton, and R. McCleery, editors. Population biology of passerine birds. Springer-Verlag, Berlin, Germany.

Vander Werf, E. 1992. Lack's clutch size hypothesis: an examination of the evidence using meta-analysis. Ecology 73: 1699-1705.

Ward, S. 1996. Energy expenditure of female Barn Swallows Hirundo rustica during egg formation. Physiological Zoology 69:930-951.

Weimerskirch, H. 1992. Reproductive effort in long-lived birds: age-specific patterns of condition, reproduction and survival in the Wandering Albatross. Oikos 64:464-473.

Wernham, C. V., and D. M. Bryant. 1998. An experimental study of reduced parental effort and future reproductive success in the puffin Fratercula arctica. Journal of Animal Ecology 67:25-40.

Westerterp, K. R., and D. M. Bryant. 1984. Energetics of free existence in swallows and martins (Hirundinidae) during breeding: a comparative study using doubly labeled water. Oecologia 62:376-381.

Wheelwright, N. T., J. Leary, and C. Fitzgerald. 1991. Costs of reproduction in the Tree Swallow (Tachycineta bicolor). Canadian Journal of Zoology 69:2540-2547.

Wiggins, D. A. 1990. Clutch size, offspring quality, and female survival in Tree Swallows--an experiment. Condor 92:534-537.

Williams, G. C. 1966. Natural selection, the cost of reproduction, and a refinement of Lack's principle. American Naturalist 100:687-690.

Winkler, D. W., and P. E. Allen. 1995. Effects of handicapping on female condition and reproduction in Tree Swallows (Tachycineta bicolor). Auk 112:737-747.

Woodard, J. D., and M. T. Murphy. 1999. Sex roles, parental experience and reproductive success of Eastern Kingbirds. Animal Behaviour 57: 105-115.
       Lack's food-limitation model for the evolution of clutch size
      and six other hypotheses that have been proposed to explain why
     birds fail to lay larger and seemingly more productive clutches.
       Hypothesis               Critical prediction
Lack's model            Food supply limits parent's ability to
                         feed young and
                         the most productive brood size equals
                         model clutch size.
Intergenerational costs Offspring quality, post-fledging
                         survivorship, or
                         recruitment dealings with increasing
                         brood size.
Intragenerational costs Parental survival or future fecundity
                         declines with increasing brood size.
Cliff edge              Fledging success and recruitment are
                         asymmetrically
                         low in broods above the modal cutch
                         size.
Parastitism insurance   Same as for cliff-edge hypothesis,
                         plus ther is
                         evidence of frequent intraspecific
                         brood parasitism.
Bad years, BYH          Productivity increases linearly with
                         brood
                         size in "good" years, but productivity
                         peaks
                         at intermediate brood size in "bad"
                         years.
Individual              For pairs beginning with the same
  optimization, IOH      number of eggs,
                         enlarged broods fledge the same or
                         fewer young than
                         broods in which the number of young in
                         the
                         nest equals the number of eggs laid.


A comparison of regression coefficients and adjusted coefficients of determination ([r.sup.2] or [R.sup.2]) describing the relationship between productivity and nestling morphological characters, assuming either that brood size carries a direct cost regardless of initial clutch size (cost hypothesis) or that the cost of rearing different-sized broods varies with initial clutch size (CS) and the degree to which the original brood size is manipulated (MBS; individual-optimization hypothesis).
                               Cost hypothesis
                                   Coeffi-
Breeding statistic          n       cient      [r.sup.2]   P
Productivity
 Proportion fledged        384     -0.402        0.000   0.294
 Number fledged            384      0.620        0.133   0.000
Nesting morphlological
  character
 Mass(g)                   251     -1.151        0.162   0.000
 Ninth primary length (mm) 251     -0.537        0.035   0.002
 Tarsus length (mm)        251     -0.067        0.009   0.090
 Bill length (mm)          211     -0.037        0.000   0.464
                           Individual-optimization
                                 hypothesis
                                     CS                    MBS
                                   Coeffi-               Coeffi-
Breeding statistic                  cient            P    cient    P
Productivity
 Proportion fledged                -0.018          0.782 -0.052  0.280
 Number fledged                     0.581          0.000  0.645  0.000
Nesting morphlological
  character
 Mass(g)                           -0.921          0.002 -1.071  0.000
 Nineth primary length (mm)        -0.834          0.005 -0.369  0.043
 Tarsus length (mm)                -0.076          0.272 -0.051  0.233
 Bill length (mm)                   0.052          0.545  0.064  0.244
Breeding statistic         [R.sup.2]   P
Productivity
 Proportion fledged          0.002   0.555
 Number fledged              0.132   0.000
Nesting morphlological
  character
 Mass(g)                     0.138   0.000
 Nineth primary length (mm)  0.035   0.005
 Tarsus length (mm)          0.002   0.307
 Bill length (mm)            0.001   0.390


Estimation of fledgling survival. (A) Results of stepwise multiple-regression analyses relating number of fledglings alive at 30 d ("Number") to brood size (BS) and the combination of clutch size (CS) and the second-order polynomial of manipulated brood size (MBS). (B) Partial correlations and power of the test relating the number of young alive to brood size, fledging date and average nestling mass, and ninth primary, tarsometatarsus, and bill lengths after controlling for the effects of brood size (cost hypothesis) or the combination of clutch size and the second-order polynomial of brood size (optimization hypothesis).
                      A) Stepwise multiple regression
                          Cost hypothesis
Equation:            Number = 0.51 + 0.59(BS)
Statistics: [r.sup.2] = 0.288, p [less than] 0.001 [++]
                      Optimization hypothesis
Equation:      Number = 0.27 + 0.71(CS) + 0.42(MBS)
                     - 0.29 [(MBS).sup.2] [+]
Statistics: [R.sup.2] = 0.321, P [less than] 0.001 [++]
                              B) Correlations
                     Partial correlation             Partial correlation
Variable                 Coefficient       P   Power     Coefficient
[(Brood size).sup.2]       -0.140        0.200 0.30         N.A.
Date of fiedging           -0.042        0.722 0.07        -0.001
Nestling mass               0.023        0.844 0.05         0.015
Primary length              0.031        0.794 0.05         0.025
Tarsus length               0.183        0.107 0.37         0.147
Bill length                 0.254        0.045 0.56         0.248
Variable                              P   Power
[(Brood size).sup.2]                 ...   ...
Date of fiedging     [greater than] 0.90   0.05
Nestling mass        [greater than] 0.90   0.08
Primary length                      0.834  0.11
Tarsus length                       0.162  0.29
Bill length                         0.054  0.52
Note:
N.A. indicates "not applicable."
(+.)P values of regression coefficients for CS (P [less than] 0.001), MBS
(P = 0.003); and MBS [2] (P = 0.020) wereall significant.
(++.)Coefficients of determination were adjusted for the number of
independent variables.
           Annual variation in productivity in relation to brood
          size (linear model) and the second-order polynomial of
                       brood size (quadratic model).
        Linear model                 Quadratic model
Year n       BS        P   [r.sup.2]       BS          P   [(BS).sup.2]   P
1989 28    1.100     0.001   0.319       -0.279      0.846     0.233    0.330
1990 36    0.606     0.084   0.059       -3.336      0.121     0.617    0.064
1991 51    0.823     0.000   0.237       -2.251      0.098     0.457    0.024
1992 36    0.108     0.754   0.001       -3.132      0.188     0.483    0.170
1993 40    0.361     0.090   0.049        2.821      0.004    -0.409    0.009
1994 54    0.622     0.002   0.156        0.920      0.405    -0.048    0.783
1995 46    0.861     0.000   0.357        1.155      0.166    -0.049    0.716
1996 48    0.847     0.009   0.122       -2.250      0.374     0.509    0.220
1997 47    0.541     0.014   0.107        1.675      0.166    -0.173    0.338
Year [r.sup.2]
1989   0.318
1990   0.126
1991   0.300
1992   0.002
1993   0.191
1994   0.140
1995   0.344
1996   0.133
1997   0.106


Notes: Values in the table are the regression coefficients relating number of fledged young to brood size (BS) or the combination of BS and [(BS).sup.2]. Sample size (n = number of broods), significance level (P), and coefficient of determination ([r.sup.2], adjusted for degrees of freedom) are given for each year.
            Results of stepwise logistic regression analyses of
        female Eastern Kingbird survival in relation to year, brood
            size (BS), clutch size (CS), manipulated brood size
        (MBS), and the square of brood and manipulated brood size.
                             [delta]
Variable        Deviance  df Deviance df  Coefficient [+] 1 SE    P
Null model       265.38  194   ...    ...       ...        ...   ...
a) Year          261.37  193   4.01    1      -0.736      0.370 0.045
b) BS            261.37  192   0.00    1       0.002      0.160 0.989
c) BS                                          2.545      1.079 0.018
  [(BS).sup.2]   255.47  191   5.90    2      -0.391      0.164 0.017
d) CS                                          0.390      0.292 0.181
   MBS           258.55  191   2.82    2      -0.161      0.181 0.372
e) CS                                          0.378      0.293 0.196
   MBS                                        -0.147      0.184 0.423
  [(MBS).sup.2]  258.31  190   3.06    3       0.080      0.163 0.623


Notes: Year effects (1992 and 1996 vs. others) are compared against the null model (a), and then separate analyses are conducted to compare the null model (with year effects included) to models that included null effects along with (b) BS, (c) both BS and [(BS).sup.2], (d) both CS and MBS, and (e) the combination of CS, MBS, and [(MBS).sup.2]. The change ([delta]) in deviance is a measure of the ability of each model to explain variation in survival in comparison to the null model.

(+.)An estimate of the regression coefficient for each variable.
                 Results of logistic regression analyses of
                male Eastern Kingbird survival in relation
               to year, brood size (BS), clutch size (CS), mani-
                 pulated brood size (MBS), and the square
                   of brood and mainpulated brood size.
                              [delta]      Regression results
Variable         Deviance df  Deviance df     Coefficient [+] 1 SE  P [++]
Null Model        249.23  192   ...    ...        ...          ...   ...
a) Year           242.95  191   6.28    1        -0.196       0.080 0.014
b) BS             242.95  190   0.00    1        -0.003       0.176 0.988
c) BS                                             0.888       1.045 0.395
   [(BS).sup.2]   242.21  189   0.74    2        -0.140       0.162 0.387
d) CS                                             0.001       0.304 0.998
   MBS            242.95  189   0.00    2        -0.004       0.207 0.984
e) CS                                             0.017       0.307 0.956
   MBS                                           -0.017       0.216 0.936
   [(MBS).sup.2]  242.78  188   0.17    3         0.086       0.211 0.684


Notes: Year effects are compared against the null model (a), and then separate analyses are conducted to compare the null model (with year effects included) to models that included null effects along with (b) BS, (c) both BS and [(BS).sup.2], (d) both CS and MBS, and (e) the combination of CS, MBS, and [(MBS).sup.2] The change ([delta]) in deviance is a measure of the ability of each model to explain variation in survival in comparison to the null model.

(+.)An estimate of the regression coefficient for each variable.

(++.)Significance of the regression coefficient.
         Future breeding performance of female and male kingbirds
          in relation to experimental modification of brood size.
                    Reduced brood         Control         Enlarged brood
Variable                Mean      I SD n   Mean   1 SD n       Mean      1 SD
Females
  Breeding date [+]      155      3.39 31   154   4.74 35      154       4.55
  Clutch size            3.5      0.57 29   3.4   0.59 39      3.4       0.50
  Addled eggs            0.1      0.28 24   0.2   0.41 38      0.3       0.48
  Brood size             2.4      1.67 24   2.4   1.54 33      2.1       1.57
  Starved young          0.2      0.54 16   0.2   0.54 21      0.4       0.74
  Fledged young          1.9      1.73 27   1.8   1.64 34      1.2       1.33
Males
  Breeding date [+]      156      4.94 34   155   4.53 54      154       3.32
  Clutch size            3.2      0.61 33   3.2   0.58 49      3.4       0.51
  Addled eggs            0.3      0.52 33   0.3   0.48 54      0.2       0.44
  Brood size             2.0      1.50 29   2.4   1.25 29      2.6       1.37
  Starved young          0.2      0.56 15   0.3   0.53 29      0.6       0.92
  Fledged young          1.3      1.47 29   1.7   1.44 51      1.5       1.42
                       ANOVA
Variable            n   F    P
Females
  Breeding date [+] 15 0.29 0.75
  Clutch size       18 0.19 0.82
  Addled eggs       15 1.92 0.15
  Brood size        15 0.31 0.74
  Starved young      8 0.28 0.76
  Fledged young     16 1.20 0.31
Males
  Breeding date [+] 18 1.74 0.18
  Clutch size       19 0.98 0.38
  Addled eggs       17 0.02 0.96
  Brood size        17 1.40 0.25
  Starved young     11 1.58 0.21
  Fledged young     17 0.58 0.57


Notes: All reductions (manipulated brood size = -l and -2 young) and enlargements (manipulated brood size = +1 and +2 young) of brood size are included under "reduced" and "enlarged" categories, respectively. Differences were compared using analysis of variance, for which the F test is reported.

(+.)Breeding date: day of year is counted consecutively from 1 January (e.g., 155 = 4 June).
               Results of experimental field studies that have
              measured interseasonal costs of reproduction for
                 adult passerines in response to brood-size
                              manipulations.
                                       Adult
                    Adult mass        Survival         Laying date
Species                 M      F         M      F          M       F
Tyrannus tyrannus       - [+]  - [++]    - [oo] - [*]      ns      + [oo]
Tachycineta bicolor            - [++]           ns
Tachycineta bicolor                      - [oo] ns                 ns
Tachycineta bicolor                             ns
Ficedula hypoleuca             - [*]     - [*]  ns
Ficedula albicollis                      ns     ns                 - [*]
Ficedula albicollis     ns     - [*]     ns     ns
Parus major                              ns     - [*]
Parus major                              ns     ns
Parus major                              ns     ns
Parus caeruleus         ns     - [*]     + [oo] - [*]      ns      ns
Parus caeruleus                          - [+]  - [oo]      + [*]  + [oo]
Parus montanus          - [+]  - [oo]    - [+]  ns
Parus montanus          ns     ns        - [+]  ns         ns      ns
Passer domesticus       - [+]  ns        - [oo] ns
Corvus frugileus                         ns     ns                 - [*]
                    Clutch       Procut-
                     size         ivity
Species               M    F       M     F
Tyrannus tyrannus     ns   ns      ns    - [+]
Tachycineta bicolor
Tachycineta bicolor        ns            ns
Tachycineta bicolor
Ficedula hypoleuca
Ficedula albicollis        - [*]         - [*]
Ficedula albicollis
Parus major
Parus major           ns   ns      ns    ns
Parus major           ns   ns      ns    ns
Parus caeruleus       ns   ns      - [*] ns
Parus caeruleus       ns   ns      ns    ns
Parus montanus
Parus montanus        ns   ns      ns    ns
Passer domesticus
Corvus frugileus           ns            - [*]
Species                        Reference
Tyrannus tyrannus   This study
Tachycineta bicolor DeSteven (1980)
Tachycineta bicolor Wheelwright et al. (1991)
Tachycineta bicolor Wiggins (1990)
Ficedula hypoleuca  Askenmo (1977, 1979)
Ficedula albicollis Gustafsson and Sutherland, (1988) [ss];
                      Gustafsson and Part (1990)
Ficedula albicollis Torok and Toth (1990)
Parus major         Boyce and Perrins (1987)
Parus major         Pettifor et al. (1988)
Parus major         Tinbergen and Daan (1990) [II]
Parus caeruleus     Nur 1984, (1988)
Parus caeruleus     Pettifor (1993b)
Parus montanus      Orell and Koivula (1988)
Parus montanus      Orell et al. (1996) [n]
Passer domesticus   Hegner and Wingfield (1987)
Corvus frugileus    Roskaft (1985)


Notes: Data for laying date, clutch size, and productivity are for survivors in the year following experimental manipulation of brood size. Results are reported separately for males (M) and females (F) if they were available. A "--" or "+" indicates that the variable varied negatively or positively, respectively, with experimental increases in brood size.

(*.)P [less than] 0.05;

(+.)P [less than] 0.10;

(++.)P [less than] 0.25;

(oo.)P [less than] 0.50,

ns = P [greater than or equal to] 050.

(ss.)Clutch size of female recruits adversely affected by brood-size enlargement.

(II.)Clutch size of female recruits unaffected by experimental increase in brood size.

(n.)Brood size manipulation did not affect timing of moult in either sex.
            The proportion of adults returning to breed plotted
             against brood size. Vertical bars indicate +2 SE,
           and numbers above each bar give sample sizes. Return
           rates of birds that failed during incubation and that
            did not renest (= failed breeders) are also given.
                      Proportion of Adults Returning
Brood Size      Males Females
Failed Breeders  10     10
Brood size
2                53     57
3                83     74
4                46     50
5                11     14
        Frequency distribution clutch size (shaded bars, right-hand
          axis scale) for initial nests of the season within the
         Charlotte Valley (Delaware County, New York, USA) Eastern
            Kingbird population from 1989 through 1997. Average
           number of young to fledge from natural and artificial
           broods of 1-5 young (black bars) are also shown; data
             are means + 2 SE. The numbers above the bars give
                               sample sizes.
Number of Eggs or Young Fledglings Clutch size
1                               10           0
2                               89          43
3                              176         312
4                               87         148
5                               24           0
COPYRIGHT 2000 Ecological Society of America
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2000 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:MURPHY, MICHAEL T.
Publication:Ecological Monographs
Geographic Code:1U2NY
Date:Feb 1, 2000
Words:16037
Previous Article:PREFERENCE-PERFORMANCE RELATIONSHIPS AND EFFECTS OF HOST PLANT CHOICE IN AN HERBIVOROUS MARINE AMPHIPOD.
Next Article:DIRECT AND INDIRECT EFFECTS OF PREDATION BY COMMON EIDERS AND ABIOTIC DISTURBANCE IN AN INTERTIDAL COMMUNITY.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |